Indoor target detection method based on improved yolov3

Abstract

The invention provides an indoor target detection method based on improved yolov3. The indoor target detection method is suitable for detection of target furniture which is small in area after being shielded in an indoor environment. The method comprises the following steps: crawling an indoor scene picture, manually labeling a detection target, and constructing a training set; clustering the target size in the training data set by using a K-means method to obtain an optimized anchor size; the method comprises the following steps of: improving a Darknet53 network of yolov3, and adding a combined module (DS module) of expansion convolution and deep separation convolution into a residual module; and performing indoor target detection by using the trained improved network. According to the method, the image information utilization rate is increased, the accuracy of indoor target detection is improved, and the detection of target furniture with a small area after shielding is also high in accuracy.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to an indoor target detection method based on an indoor data set. Background technique [0002] Target detection technology has become a widely used technology in the field of computer vision at this stage. Traditional target detection is mostly based on manually labeled features, and its accuracy is poor. With the development of the times, it has gradually failed to keep up with people's requirements for accuracy and efficiency. With the upsurge of artificial intelligence such as the development of computer vision, through deep learning networks such as convolutional networks, image features can be extracted by computer to complete target detection tasks with high precision, and its speed, accuracy and robustness Both have improved. At present, the target detection method based on the convolutional neural network has become the mainstream. [0003] The applic...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problem of poor indoor target detection accuracy caused by poor target learning and mutual occlusion of indoor targets when the yolov3 target detection network detects indoor objects, this invention proposes an indoor target detection method based on improved yolov3. The network is improved to improve the detection accuracy of indoor objects

Images